history data

[faridseifi]

• Use May data to generate and capture history for 6 months
  • when history is not used by pipeline
  • when history is fully used by pipeline
  • when history is used by pipeline but the between-causal-pathway MPM weight for history is not used
• Using the 3 spread sheets resulting from the item above, compare average (normalized) count of messages (causal pathways and measures) per provider and see the effect of history.
• Compare population statistics with and without history. Histogram of counts of messages of each message type, each causal pathway and each measure with and without history.
• add passed rate to the history extracted for each row
• Update the readme with the new extract history script information
• (Not now) Compare the effect of turning on only history versus no scoring at all. Also try preferences only then add history then MI and see the effect.